The purpose of the present Perspectives is to present a synopsis of the literature on bacterial “quorum sensing” as a background for the proposal that interference with this communication system offers potential targets for the design of novel antibiotic drugs. Quorum sensing is the recently discovered chemical communication system among bacteria (both Gram-positive and -negative). It is vital for intra- and interbacterial gene regulation and for keeping bacterial colonies (“biofilms”) intact, allowing resident bacteria to assume specialized roles that contribute to enhanced survival of the group. There are several processes involved in quorum sensing that are familiar to pharmacologists; i.e., specific signaling molecules bind to and activate receptors that transduce the quorum-sensing signal into intracellular second messenger responses. We highlight herein the similarity between quorum-sensing communication to ligand-receptor interactions, suggesting that inhibitor drugs could be designed using current standard pharmacologic principles. Such drugs would have novel mechanisms of action and might therefore be more effective against antibiotic-resistant strains of bacteria.
When admitted to a hospital, one expects to be treated for the presenting condition, not incur a new one. Yet, every year, an estimated two million people acquire nosocomial infections (Weinstein, 1998) that may be more difficult to treat because many bacteria are resistant to at least one antibiotic and some are resistant to all commonly used antibiotics. For many years, vancomycin provided a last resort against treating resistant Gram-positive infections, but there are now reports of vancomycin-resistant strains (Lowy, 2003). Unfortunately, the development of antibiotic resistance continues to outpace the development of new antibiotics (Walsh, 2003).
Multiple factors contribute to resistance, including overuse, infections in immune-compromised patients, and increased use of indwelling medical devices, which provide a fostering environment (Donlan, 2002). The prevalence of biofilms (a strongly adherent assemblage of differentiated microbial cells enclosed in a matrix of polysaccharides) (Stoodley et al., 2002) in infections and on surfaces of medical implant devices has focused attention on the increased antibiotic resistance (1000-fold) of biofilm-resident bacteria versus the more commonly studied planktonic (free-floating) form. It has recently been suggested that biofilm-resident bacteria “communicate” by a process termed quorum sensing and that this contributes to their competitive advantage and enhanced antibiotic resistance. Quorum sensing, which is the detection of the surrounding cell density and activation of appropriate compensatory regulation of cell function, uses chemical signaling and “sensor” molecules. Described in an early review by Fuqua et al. (1994), quorum sensing has subsequently been found to be widespread in Gram-positive and -negative bacteria (Sturme et al., 2002). In this process, compounds diffuse from, or are secreted from, bacteria as the population grows. These compounds—such as γ-butyrolactones and “auto-inducing peptides” (autoinducers) in Gram-positive bacteria and N-acyl homoserine lactones, quinolones, or cyclic dipeptides in Gram-negative bacteria (Hastings and Greenberg, 1999; Dunn and Handelsman, 2002)— diffuse away from the cell and interact with the same or other cells by attaching to and activating specific cell surface-associated or intracellular receptors. Once sufficient signal is detected, transduction leads to the induction of genes that control a variety of survival functions, including the production of antimicrobial substances and protection against the host's defense mechanisms (Salmond et al., 1995). The recent discovery of an autoinducer produced by the luxS gene found in Gram-positive and -negative bacteria suggests the possibility of “cross-talk” between the two bacteria types (Dunn and Handelsman, 2002).
Several organisms seem to have evolved the ability to interrupt this process. Examples include plants (e.g., tomato, rice, and pea) and soil bacteria that secrete compounds that alter homoserine lactone activity and Delisea pulchra, which secretes a halogenated furanone that inhibits quorum-sensing signaling (Bauer and Robinson, 2002). This suggests that synthetic analogs of such substances, or novel compounds from drug discovery efforts, could interrupt quorum sensing in one or more (Stewart, 2003) ways.
It is the purpose of this review to highlight the similarity of quorum-sensing processes to ligand-receptor binding and the use of this construct as a guide to direct novel antibiotic drug design efforts based on standard pharmacologic principles and drug discovery processes. The unique nature of their mechanism should provide these new antibiotics with greater activity against currently resistant bacteria.
Biofilms have been perhaps most succinctly described as “an assemblage of microbial cells that is irreversibly associated (not removed by gentle rinsing) with a surface and enclosed in a matrix of primarily polysaccharide material” (Donlan, 2002). Images of biofilms can be found at http://www.microbelibrary.org (search for “biofilm”) (Fig. 1). Bacteria can adhere to surfaces and form biofilms in places as seemingly diverse as teeth, lungs, intestines, contact lenses, and water pipes. Biofilm formation is considered the major contributor to the virulence of Staphylococcus epidermidis, the most frequent cause of nosocomial sepsis and catheter-related infections (Vuong et al., 2003).
Many bacterial species are now known to form biofilms (Donlan, 2002), and Gram-positive and -negative bacteria and yeasts can coexist within the same biofilm. The bacteria in biofilms can be differentiated from free-floating planktonic forms by an extracellular polymeric substance, slower growth rate, and the up- or down-regulation of certain genes. The extracellular polymeric substance acts as a filter and conduit for nutrients and minerals that are channeled to interior cells and protects cells from potentially harmful agents, including antibiotics (Donlan, 2002). Furthermore, plasmid exchange occurs faster in biofilms because of greater cell-to-cell contact (Donlan, 2002). It is postulated that biofilms contribute to antibiotic resistance by at least three mechanisms: reduced antibiotic penetration across the extracellular polymeric substance; a favorable (e.g., anaerobic) environment within the inner layers; and bacteria cell differentiation and specialization of function that provides increased protection (Stewart and Costerton, 2001).
Biofilm-growing bacteria can become resistant to antibiotics sooner than their planktonic counterparts due to protective features of the film such as impedance against diffusion and favorable environment within the film (Stewart and Costerton, 2001). Even if antibiotic therapy is effective against some of the colony, surviving bacteria can feed off nutrients left behind (Costerton and Stewart, 2001). Natural defenses against bacteria, such as mucosal secretions, are less effective once biofilms have formed (Singh et al., 2002). As a result, bacteria in biofilms survive exposure to concentrations of antibiotics 1000-fold greater than are lethal when the cells are in suspension (Stewart and Costerton, 2001).
Biofilms develop on the insides of distribution pipes, and rogue bacteria eventually break off and enter the water supply. Conventional disinfectant techniques such as chlorination often are ineffective. In medical settings, biofilms occur in a myriad of places, such as the intestinal brush border (e.g., Vibrio cholerae), urethra lining (e.g., Neisseria gonorrhoeae), lymphoid patches in the intestine (e.g., Salmonella typhimurium) (Costerton et al., 1978), antibiotic-recalcitrant acne (Coates et al., 2003), chronically infected tonsils (Chole and Faddis, 2003), cystic fibrosis (lungs) (Prince, 2002), urinary and central venous catheters, and mechanical heart valves (Donlan, 2002).
Description and Overview
Quorum sensing is a form of bacterial communication that helps regulate group behavior. Bacteria release chemical substances called autoinducers into their surroundings. As the population density increases, so does autoinducer concentration. When the population density is sufficiently high (i.e., a quorum is achieved), autoinducer concentrations become high enough to bind to receptors on/within the source or nearby bacteria. The signal is then transduced into an intracellular biochemical signal or altered gene expression in the target bacteria. This induces a variety of adaptive physiological changes such as bioluminescence, production of antibiotics, and activation of biofilm formation. Quorum sensing is widespread. For example, it occurs in squid light organs (e.g., Vibrio fischeri) (Visick and McFall-Ngai, 2000), soil and plant roots (Dunn and Handelsman, 2002), and human infections (e.g., Pseudomonas aeruginosa in cystic fibrosis, ocular infections, and burns) (Rumbaugh et al., 2000). It occurs in Gram-positive (e.g., Streptococcus pneumoniae, Bacillus subtilis, and Staphylococcus aureus) and -negative (e.g., V. fischeri, P. aeruginosa, Agrobacterium tumefaciens, and Erwinia carotovora) bacteria and is a mechanism of cross-talk between Gram-positive and -negative bacteria (Miller and Bassler, 2001; Dunn and Handelsman, 2002). One autoinducer, N-octanoyl-l-homoserine lactone (AI-2), is proposed to be a signaling molecule in all bacteria (Xavier and Bassler, 2003) that is essential for the formation of mixed-species biofilms containing Porphyromonas gingivalis and Streptococcus gordonii (McNab et al., 2003).
Autoinducers and Their Receptors
Three types of autoinducers have been identified to date: acylated homoserine lactones (AHLs, acyl-HSLs, or HSLs), such as N-3-oxohexanoyl-l-homoserine lactone (AI-1), which are found in Gram-negative bacteria; autoinducing peptides (AIPs), which are found in Gram-positive bacteria; and autoinducer-2 compounds (AI-2s), which are found in Gram-negative and Gram-positive bacteria.
At least 25 species of Gram-negative bacteria (excluding Vibrio harveyi and Myxococcus xanthus) use “LuxI/LuxR-type” quorum sensing similar to that used by V. fischeri. LuxI, an AHL synthase, is involved in the biosynthesis of AHL from fatty acids. LuxR is an AHL-dependent transcriptional regulatory protein. AHL binds in a dose-related manner to the LuxR “receptor” that is localized to the cytoplasm and cytoplasmic face of the inner bacterial cell membrane. The AHL-LuxR complex activates target gene transcription through intracellular biochemical pathways (Kolibachuk and Greenberg, 1993; Miller and Bassler, 2001; Whitehead et al., 2001) (Fig. 2A). Many other species of Gram-negative bacteria use a similar chemical signaling system.
AIPs are amino acids or short peptides synthesized in Gram-positive bacteria and are processed, modified, and exported by the ATP-binding cassette export systems (Kolibachuk and Greenberg, 1993; Sturme et al., 2002). AIPs bind to cell surface-bound histidine protein kinase, which autophosphorylates and in turn phosphorylates a response regulator that activates transcription of one or more target genes (Miller and Bassler, 2001; Sturme et al., 2002) (Fig. 2B).
AI-2s, common to both bacteria types, are derived from furanones. The specific structures are yet to be determined. An exception is a furanosyl cyclic borate diester recently identified as the AI-2 secreted by V. harveyi (Chen et al., 2002). This AI-2 is encoded by the luxS gene. It binds to a LuxP protein (a LuxR homolog) (Coulthurst et al., 2002). The AI-2/LuxP complex then binds to membrane-bound histidine protein kinase, and signal transduction occurs by multistep phosphorylation similar to that of AIPs (Miller and Bassler, 2001; Xavier and Bassler, 2003). In other bacteria, extracellular AI-2 is transported back into the cell through a LuxS-regulated (Lsr) transporter (Taga and Bassler, 2003) (Fig. 2C).
Intracelleular Signal Transduction
Autoinducer-induced transcriptional changes include protective bioluminescence, increased virulence, biofilm formation, antibiotic production, increased competence, and sporulation. For example, V. fischeri produces a mutually beneficial camouflaging bioluminescence inside squid (Whitehead et al., 2001); quorum sensing in P. aeruginosa contributes to biofilm formation in persistent lung infections of cystic fibrosis patients and eye infections caused by wearing contact lenses (Costerton et al., 1999); deletion of one or more quorum-sensing genes in P. aeruginosa decreases the bacterium's destructiveness and mortality rate in burn wound infections (Rumbaugh et al., 1999); a mutant strain of P. aeruginosa has a quorum-sensing gene that produces a biofilm sensitive to biocide, a treatment to which biofilms are normally resistant (Davies et al., 1998); autoinducers are found in P. aeruginosa biofilms growing on urethral catheters (Stickler et al., 1998); AI-2 is required for the mixed-species biofilms of P. gingivalis and S. gordonii in dental plaque (McNab et al., 2003); and quorum sensing at high cell densities enhances the competence of soil bacteria B. subtilis by enabling the bacteria to take in exogenous DNA from cell lysis and using it to form dormant spores during nutrient-deprived periods or to repair damaged or mutant chromosomes (Miller and Bassler, 2001). A particularly intriguing example is the transcriptional change in E. carotovora (potato-rotting bacteria), resulting in production of carbapenem, a β-lactam antibiotic that kills competing bacteria (Axelrood et al., 1988).
Biofilms and Quorum Sensing as Antibiotic Targets
Multiple approaches, including nonsteroidal anti-inflammatory drugs (Alem and Douglas, 2004), are being investigated for use in attacking biofilm structure or integrity or optimizing treatment using currently available antimicrobial agents. The present review focuses on interference with quorum sensing.
The premise of the present Perspectives is that quorum-sensing signaling can be interrupted in similar manners to ligand-receptor pathways, i.e., by inhibiting ligand synthesis, transport, or release; inhibiting receptor synthesis and processing; and perhaps most analogous to current pharmacotherapy, inhibiting enzyme activity or ligand-receptor binding.
Targets for Drug Design: 1. Receptors
The currently identified receptor targets are those for AHLs, AIPs, and AI-2.
AHLs. The receptors for AHLs are the LuxR family of transcriptional regulators (Hanzelka and Greenberg, 1995), localized to the cytoplasm and cytoplasmic face of the inner membrane. Interaction of AHL with the N-terminal region of LuxR unmasks LuxR's C-terminal DNA-binding domain (which is blocked/inhibited by the N-terminal domain). The AHL-LuxR complex binds to specific promoters and activates transcription (Fuqua et al., 2001).
AIPs. The receptors for AIPs are located on the cell membrane and comprise 5 to 8 transmembrane segments in their N-terminal domain and a common histidine protein kinase-type C-terminal domain (Kleerebezem et al., 1997). Upon AIP binding, the receptor kinase is activated, leading to its autophosphorylation. The activated receptor then phosphorylates the response regulator, which in turn activates several genes, including the genes for AIP, the receptor, the ATP-binding cassette exporter, and the response regulator.
AI-2. The target receptor for AI-2 is species-dependent (Taga and Bassler, 2003). For example, in V. harveyi it is LuxP, a homolog of LuxR. In S. typhimurium, extracellular AI-2 binds to the Lsr transporter and is internalized into the cell, where it acts on AI-2-regulated genes.
Targets for Drug Design: 2. Signaling Molecules
The quorum-sensing signaling molecules thus far identified fall into distinct chemical families, broadly classified as acylated homoserine lactones (found in Gram-negative bacteria), oligopeptides (found in Gram-positive bacteria), and AI-2s (found in both types) (Taga and Bassler, 2003). Each of these could be used as chemical template starting points for drug discovery efforts using standard high-throughput screening or molecular modeling approaches. Some of these templates include the following.
AHLs.N-Acyl-homoserine lactones (Fig. 3A–E), first identified in V. fischeri (Whitehead et al., 2001), vary in the size (4–14 carbon atoms) and composition (double bonds or hydroxyl groups) of the acyl chain. Over 50 species of bacteria are known to use AHLs for quorum sensing (Fuqua et al., 2001). AHLs are also called AI-1s and seem to be used exclusively for intraspecies communication in Gram-negative bacteria. 2-Heptyl-3-hydroxy-4-quinolone is an autoinducer found so far only in P. aeruginosa, which uses two quorum-sensing pathways. It is a regulatory link between the two pathways (Miller and Bassler, 2001).
AIPs. AIP autoinducers are peptides post-translationally modified to yield a variety of diverse structures (Sturme et al., 2002) (Fig. 3F) that provide binding selectivity and signal specificity. Unlike the AHLs, which freely diffuse out of the cell, AIPs are actively secreted. AHLs can be linear and have dehydrated amino acids, an N-terminal extension that is removed during or after secretion (Kleerebezem et al., 1997), or specific features such as the cyclic lactone in Enterococcus faecalis or the cyclic thiolactone in S. aureus (Sturme et al., 2002) (Fig. 3G). Interestingly, an AHL recently identified in V. harveyi produces a signal similar to that in Gram-negative bacteria, but the response is similar to that in Gram-positive bacteria (Xavier and Bassler, 2003).
Targets for Drug Design: 3. Disruption of Quorum Sensing by Plants and Bacteria
Some species of plants and bacteria have evolved chemicals that disrupt the quorum sensing of other species. The marine red alga D. pulchra secretes halogenated furanones and enones that structurally resemble the AHL autoinducer of Serratia liquefaciens, antagonize the binding of AHLs at the receptor, and successfully ward off infestation of S. liquefaciens by interrupting quorum-sensing-mediated swarming motility (Miller and Bassler, 2001). Tomatoes, peas, soybeans, rice, crown vetch, and Medicago truncatula (a legume closely related to alfalfa) secrete chemical compounds (not yet isolated) that stimulate specific AHL receptors, and compounds from alfalfa root stimulate the production of the antibiotic zwittermicin A by Bacillus cereus, which inhibits the more pathogenic Pythium torulosum (Dunn and Handelsman, 2002). Strains of S. aureus gain competitive advantage when their AIPs inhibit the quorum sensing of other strains (Miller and Bassler, 2001). This is an example of intraspecies interference. An example of interspecies interference occurs in the case of the soil bacteria B. subtilis and E. carotovora. B. subtilis secretes the enzyme AiiA (aiiA gene product), a homolog of zinc-binding metallohydrolases. AiiA inactivates the AHL autoinducer of E. carotovora (by hydrolyzing either the amide bond between the acyl side chain and the lactone ring of the AHL or the ester bond within the lactone ring) (Dong et al., 2000), rendering it inactive. There is even a strain of bacteria (Variovorax paradoxus) that disrupts the quorum sensing of other bacteria by ingesting and using their quorum-sensing compounds as a nutrient source of carbon and nitrogen (Leadbetter and Greenberg, 2000). Interruption of quorum sensing impedes biofilm formation of Escherichia coli (Ren et al., 2001). These natural molecules could serve as chemical templates for drug discovery efforts using standard high-throughput screening or molecular modeling methodology.
Design of Quorum-Sensing Antagonists
Quorum-sensing pathways bear a striking similarity to mammalian signaling pathways and, therefore, should be amenable to pharmacologic manipulation in ways similar to those used to design drugs. Autoinducer ligands interact with quorum-sensing receptors in a concentration-dependent manner, consistent with traditional concepts of receptor binding (e.g., Hanzelka and Greenberg, 1995). Also consistent with traditional concepts of receptor binding, autoinducer ligands that act at the same quorum-sensing receptor compete with each other. For example, a series of structural analogs of the P. aeruginosa autoinducer N-3-oxo-dodecanoyl homoserine lactone compete with autoinducer binding to LasR and have varying degrees of agonist activity (Passador et al., 1996), and analogs of the autoinducer 3-oxohexanoyl HSL competitively antagonize the binding of the autoinducer in V. fischeri (Schaefer et al., 1996).
Reference to Fig. 2 reveals multiple sites to target for the design of anti-quorum-sensing, anti-biofilm antibiotics. Some of these include LuxI, LuxR, the AIP receptor, LuxS, and the Lsr transporter. For Gram-negative bacteria, enzyme inhibitors of LuxI or antagonists of the LuxR receptor-binding site should be sought (Fig. 2A). For Gram-positive bacteria, antagonists of AIP processing or the AIP receptor should be sought (Fig. 2B). Enzyme inhibitors of LuxS or inhibitors of the Lsr transporter (Fig. 2C) could have application to both Gram-negative and -positive bacteria. Combinations of mechanisms would be expected to be more effective than single-mechanism approaches. Examples of some early efforts toward the design and development of such agents are summarized below.
Based on the observation that the fronds of red alga growing in Botany Bay, Australia are rarely covered with biofilms despite thousands of bacterial species in the waters, Rice et al. (1999) determined that the alga secretes halogenated furanone compounds that prevent biofilm formation and help disrupt existing biofilms (Costerton and Stewart, 2001). Hentzer et al. (2003) used gene microarray technology and transcriptome analysis to show that the targets of a synthetic furanone were genes of the quorum-sensing system of P. aeruginosa and that the compound inhibited bacterial activity in vitro and in vivo. That the effect was specific to biofilms was shown by the lack of activity against planktonic cultures of the same bacteria.
P. aeruginosa, a common cause of nosocomial infections and responsible for chronic lung infection in an estimated more than 90% of cystic fibrosis patients, uses a quinolone signal as part of its quorum-sensing system. Calfee et al. (2001) determined that anthranilic acid is a precursor to the quinolone signal and that an anthranilate analog (methyl anthranilate) inhibits the quinolone signal production and decreases the expression of cellular virulence factors in a dose-dependent fashion.
There is recent evidence that joint use of a ligand that disrupts quorum sensing and a more traditional antibiotic results in a synergistic antibacterial effect. Balaban et al. (2004) reported that the RNA III-inhibiting protein, a heptapeptide that inhibits staphylococcal biofilm formation by obstructing quorum-sensing mechanisms, and DD(13), K(4)-S4(1–13)(a), a 13-residue dermaseptin derivative believed to kill bacteria by disrupting membranes, act in synergy by attacking bacteria simultaneously via two different mechanisms.
Quorum sensing is a bacterial chemical communication system that involves signals, signal sensors, and signal transduction mechanisms. The viability of biofilms, coordinated colonies protected by an outer matrix that confers increased antibiotic resistance over planktonic bacteria, is an example of an antibiotic-resistant system that depends on the fidelity of quorum sensing and other communication systems. Interference with quorum sensing should disrupt biofilm-protective measures and greatly enhance the susceptibility of bacteria to antibiotic drugs. The nature of quorum-sensing communication—involving enzymes, transporters, and ligand-receptor interactions—presents classic pharmacologic targets for drug discovery efforts using standard techniques. The novel nature of their mechanisms of action might allow these new quorum-sensing inhibitors to be effective against bacterial strains that are currently antibiotic-resistant.
We thank Dr. Jo Handelsman (University of Wisconsin, Madison, WI) for critical review of the manuscript.
- Received July 28, 2004.
- Accepted October 25, 2004.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: AI-2, N-octanoyl-l-homoserine lactone; AHL/acyl-HSL/HSL, acylated homoserine lactone; AI-1, N-3-oxohexanoyl-l-homoserine lactone; AIP, autoinducing peptide; Lux, luminescence (lux) gene; Lsr, LuxS-regulated.
- The American Society for Pharmacology and Experimental Therapeutics